As of this date and time in major league baseball there are only 24 players with 20 or more saves. In contrast, there are 39 with 20 or more saves+holds. There are 31 players with 10 or more saves. There are 74 with 10 or more saves+holds.

The top 10 in saves+holds is reassuringly competent:

37 Heath Bell (37 saves)
35 Rafael Soriano (35 saves)
35 Joakim Soria (35 saves)
35 Brian Wilson (35 saves)
34 Francisco Cordero (33 saves, 1 hold)
33 Neftali Feliz (30 saves, 3 holds)
32 Luke Gregerson (1 save, 31 holds)
31 Matt Capps (32 saves)
30 Jonathan Papelbon (30 saves)
30 Billy Wagner (30 saves)

Time to consider a change.

First, head-to-head is not rotisserie. ERA and WHIP are incredibly volatile on a week-to-week basis. Far more predictable are strikeouts, saves, and wins (see my Wins by Category analysis). Saves are a special category of their own (see below). But any fantasy manager can prioritize strikeouts. And any manager can examine the depth chart of run-scoring teams to find decent, if not great, pitchers that get some strikeouts and compete for wins. By this standard, Tim Hudson is wildly over-rated for head-to-head play. His prospective run support is good, but not great, and despite his sterling ERA and WHIP, he has only 27 Ks in 64.1 innings pitched. I’d rather gamble on Ian Kennedy.

Second, most leagues, rotisserie and head-to-head, use some kind of start or inning limits. In leagues with deeper rosters this often leaves room for quality non-closing relievers such as Daniel Bard or Luke Gregerson, neither of whom makes ESPN’s top 300. Both of these pitchers have more strikeouts than Tim Hudson in half the innings pitched. They help push down ERA and WHIP. Most importantly, these guys are generally available to fill out a 23- or 24-man roster. They cost little to nothing and in leagues that use a start limit, they add strikeouts, quality innings, and vulture wins and saves while subtracting no starts.

Finally, let’s consider closers. I believe that most analysts undervalue closers. In head-to-head, this problem is magnified.

The disdain for closers arises from the all-or-nothing nature of the save stat. If a high-priced closer gets injured or replaced you lose almost all of your investment. Even coupling a closer with his backup doesn’t necessarily protect you. The backup is simply not likely to have the talent or the job security of the original closer. (I know this well. I owned Eric Gagne and Yhency Brazoban in 2005.)

However, the flip side of this risk is high reward. The limited number of closers makes saves the single most predictable stat category in head-to-head. While a team of middling hitters can often take a few hitting categories from a superior hitting team, and a team of middling starting pitchers can easily make a run at the ERA and WHIP points,  the team with three or four closers will almost always take the saves point. If you go into a single elimination playoff game against an opponent with more closers, you are likely down a point before the match even starts.

High-risk goes with high-reward. The former fact should not obscure the latter.

After the top 50 players or so — the ones desired in any format in any matchup — the game being played makes all the difference. Here are three examples that focus on hitting. I’ll write about pitching in another post.

First, player strengths in head-to-head formats are different than in rotisserie formats. When stats zero out every week against a new opponent, batting average or OBP are far less important than runs, home runs, RBIs and steals (see my Wins by Category analysis). Hitters must have power or speed or both to make a significant difference in head-to-head.

Second, roster depth matters. If a league offers deep enough rosters for a manager to platoon, David Ortiz, for one, is much more attractive. Even as his terrible April has turned into a very good May, he is most potent and most likely to get at bats against right handed pitchers. A platoon of Ortiz and any comparable hitter will give you just as much value as a single slugger ranked twice as high — one, it is important to remember, that you would not be platooning.

Finally, the rankings are based on projected player performance for the rest of the season. This is the nature of the exercise. The results tell you who to own, not who to play. But most leagues allow a substantial number of transactions, making who to play right now a decisive question. If you can play a great player for a month, then get an adequate substitute off the waiver wire, you need to do it.

This is the case among catchers. There are a handful of great hitting catchers and all the rest are interchangeable. So it is bizarre to see Mike Napoli ranked so low. Since starting catcher Jeff Mathis went on the DL, Napoli has not only hit extremely well, with a 1.106 OPS in May, but has played almost every day. Angels Manager Mike Sciosa went from hating the guy to running him into the ground, and Napoli has done nothing but hit. The rub against Napoli is that he will get bumped into a backup roll again with Mathis’ return. This may be true. But it shouldn’t obfuscate the fact that at this given moment Napoli is one of the top fantasy catchers in the game. He’s up there with Joe Mauer. In head-to-head leagues where batting average or OBP is less important (see above) he is, quite simply, the best.

Now, I’m not here to tell you that any of those guys will be major fantasy forces for the rest of 2009. But if you jumped early, for instance, onto the [Scott] Richmond bandwagon, how happy are you right now?

Well, duh. With this kind of modeling, Harris could be a global warming expert.

But rather than mock the evolving rankings of ESPN’s experts — they write to meet demand, after all — I thought I’d zoom out a level above rank and talk about strategy.

There are two kinds of starting pitchers — those you can count on and all the rest.

Those you can count on are Johan Santana, Tim Lincecum, Roy Halladay, and a few others. I’d mention Brandon Webb except that he’s injured, which means you couldn’t. These are the pitchers who almost always post a quality start. A pitcher can’t hit an ERA under 3.5 without great consistency.

Then there are the rest. Even if you avoid the truly horrible (think the Texas Rangers) or the horribly erratic (think Oliver Perez), the returns are still unpredictable. Most good but not great pitchers will have numerous bad starts. And you don’t know when those bad starts will happen.

For example, Javier Vasquez has pitched well this season, posting a 3.89 ERA in seven starts through May 9. But he also gave up 5 runs and 6 runs in successive starts on April 29 and May 4.

Given that reality, I look for pitchers that couple a respectable ERA with some additional tangible fantasy advantage. Strikeouts is the most obvious. Vasquez, for example, racked up 16 Ks in 14.2 IP in the two games mentioned above. The three other factors I look for are innings, team run production, and the backing of a good bullpen. All of these are indicators of whether or not your average pitcher can stay in games and get wins.

Finally, there’s one great intangible — youth. You never know when a young pitcher will take a step up in capability. That’s what you have to hope for with Richmond, even if he is already 29.

Peripheral statistics generally come from analysis of how a ball is put in play (or not) in relation to the outcome of the play. Perhaps the best known is batting average on balls in play, or BABIP, which measures “the number of batted balls that safely fall in for a hit (not including home runs).” BABIP averages are compiled for ground balls, line drives, and any number of other splits.

BABIP is most often invoked to estimate luck. A batter hitting lots of line drives and not getting hits is supposedly due for a positive correction. Likewise a pitcher getting ground balls and not getting outs.

Thus, Tristan Crowcoft recently used two BABIP averages to tell owners not to give up on Kevin Slowey:

Kevin Slowey’s BABIP numbers very much support his case as a big-time buy-low candidate. Both his .357 BABIP on ground balls and .833 BABIP on line drives were noticeably higher than the league averages…

Going many times better, A. J. Mass processes so many peripheral stats in his recent Hit Parade column that he can’t even reveal what they are:

Next, we take the velocity [read the column] and tweak it according to a complex formula based on each hitter’s skill set, determined by a combination of stats that measure patience at the plate, ability to put the ball in play, power potential and speed.

If he told you the formula he’d have to kill you.

How much do I care about BABIP? Not much.

First, these analyses often reveal the obvious. I’m keeping Kevin Slowey because he pitched well last year and because he has sterling minor league numbers, not because he has good peripherals.

Second, these analyses rely upon the premise that the player will continue to produce (or give up) the same types of hits. Your slumping star’s line-drive BABIP may return to the league average over time, but that means nothing if your slumping star hits fewer line drives. The correction you expect won’t have any effect. The market can stay irrational longer than you can stay solvent.

I’ve tabbed a few long plays already this season. I drafted John Smoltz and slotted him to the DL. Actually, that hardly qualifies. The DL-stash of a 22nd round pick is no hardship. I drafted Pablo Sandoval on the hopes that he gains catcher eligibility at some point. That has cost a little more, as Sandoval has started slow and will be no better than an average corner infielder even when he gets better.

Now here’s two more to think about.

First, Ben Sheets. Milwaukee’s former ace could be back in August, but he’s not currently signed by any major league team, which means he’s not DL-eligible. If you sign him, you use up a roster spot for nothing. Once Sheets gets a contract, the run will be on. You need to sign him just before that happens. For now, I’m laying off.

Second, Matt Wieters. The number one prospect in all of baseball is ranked by ESPN as the 10th best catcher and the 179th pick overall. But his eventual callup to the majors still looks a month or two away, especially with the minor but worrisome hamstring strain he suffered this week. In the Mynci, Wieters just landed on the waiver wire. For me, this is a long play worth the intervening zeros. Wieters is in. Sandoval is out.

The start limit has always had a loophole. The day you use your last start, all starts count. If you go into Sunday with six of seven starts accounted for, nothing prevents you from pitching multiple Sunday starters. With four SP roster slots a manager can easily hit 10 starts in any given week. This is most problematic in head-to-head playoffs when managers must win or go home.

The result is to devalue pitching even more than it already is. By churning, a manager can win two of the five common pitching stats — wins and strikeouts — with no investment in quality starters.

The obvious solution is to assign a depth chart rating to each pitching slot. On the day the start limit is reached, the depth chart determines the cut-off between starters who count and starters who don’t.

Instead of implementing the obvious solution, ESPN makes things worse. This season the default, maximum, and minimum start limits are based on the number of pitchers on your roster. Not the number of starters. The number of pitchers. If you have nine pitching slots, the starts limit is nine (most weeks).

I am solving this problem by lowering the number of pitching slots to seven –  three starting pitchers, four relief pitchers. The two slots removed are added to the bench. This gives managers much more flexibility and changes the ESPN minimum to six for most weeks. I will use seven. Lowering the number of starter slots also limits the last-day-loophole to the lower number of starters.

The disadvantage of this change is that it forces managers to become more active in handling their starters. If they miss a few days they may easily miss some starts. The very reason I have set more starting pitching slots in the past was to make it easier for managers to set their starters by the week.

In general, the previous indicators held true. Saves and Steals offered, by far, the most predictable correlation between talent and points, followed by the other cumulative categories: Strikeouts, Wins, Runs, Home Runs, and RBIs (in that order). Then came the three percentage-based stats: ERA, OBP, and WHIP.

Pitching wins is problematic. Weeks 12-22 saw a much stronger correlation of talent to points than weeks 1-11. However, the trendline is skewed by the excellent performance of the top two teams. I recall that the top team in total pitching wins (12 of 12) used a streaming strategy to routinely reach eight or nine starts. My suspicion is that correcting for starts would increase the unpredictability of the category and bring the trendline down.

For the season, the correlation of talent to points maps as follows (trendline slope in parenthesis):

Saves (y = 0.0661x)
Steals (y = 0.058x)
Runs (y = 0.047x)
Strikeouts (y = 0.0434x)
Home Runs (y = 0.0428x)
RBIs (y = 0.035x)
Wins (y = 0.0318x)
ERA (y = 0.0223x)
OBP (y = 0.0191x)
WHIP (y = 0.0099x)

Here are the charts for weeks 12-22:

Here are the charts for the full season — the average of the charts above with the charts from weeks 1-11:

So we’re in the first week of the playoffs and I’m down 7-2-1.

That’s why, with most of my hitters sitting for today’s travel day, I signed Nick Punto, Chase Headley, Adam Laroche, and David DeJesus, giving up on Max Scherzer and the back of my bullpen in exchange. Survive and advance.

And yet, some players performing perfectly well through August will tank the rest of the year. And some players who tanked early — perhaps even Paul Konerko — will recover their stuff in September. The question is which ones.

This week I took a flyer on Fred Lewis, letting go Adam Lind who had replaced Kelly Johnson. In this process I gave up a good backup second baseman, then a great hitting prospect who seems primed to produce, and settled on a base stealer. In mind was my Wins by Category analysis of the first half. In the upcoming playoffs, currently with the top seed, I need five points a round. No more than that. Steals are worth a flyer.